Pupillary instability reflects alterations in autonomic nervous system activity and has been shown to reflect change in alertness. However, the extent to which it can predict subsequent performance impairment and alertness failure is not clear. Eighteen healthy young adults (group age = 21.44 ± 3.24 years, 10 men) underwent 40 hr of continuous wakefulness, completing an 11-min Pupillographic Sleepiness Test (PST), the Karolinska Sleepiness Scale and a 10-min Psychomotor Vigilance Task (PVT) every 2 hr. Waking electroencephalography was recorded continuously and scored for microsleeps and slow eye movements (SEMs) during PVTs. Pupillary instability was sensitive to time awake, significantly increasing after 18 hr of wakefulness. The time course of impairment was almost identical to PVT lapses, microsleeps and SEMs. Receiver operating characteristic curve analysis demonstrated reasonable sensitivity and specificity of pupillary instability in correctly classifying PVT lapses, microsleeps and SEMs above individual baseline thresholds (all AUC values >0.78, p < 0.0001). Preliminary cut-off scores ranging from 10 to 11.5 mm/min for varying impairment thresholds are proposed for young adults. If reproducible in field settings, the PST may be a strong candidate as a fitness for duty/fitness to drive tool for detecting drowsiness-related impairment.